OPEN FAU Online-Publikationssystem der Friedrich-Alexander-Universität Erlangen-Nürnberg
Not a member yet
23478 research outputs found
Sort by
Neue (Hetero-)Arenring-Geankerte, Tripodale Arylchalkogenid-Liganden und deren Koordinationschemie mit Uran
No full textUnderstanding catalytic systems and improving upon them is of paramount importance for a more efficient conversion of substrates to products of commercial and industrial relevance. Previous research on U(III) complexes featuring mesitylene-anchored, 1,3,5-tris-phenolate ligands revealed a prolific redox and (even electrocatalytic) small molecule activation chemistry. In this work, the alteration of the chelator’s electronic properties was pursued by opportune modifications to the organic framework design. Three different approaches were deemed promising:
a. Substitution of the arene anchor’s methyl groups for substituents with increased electron-donating (EDG) or electron-withdrawing character (EWG) (Chapter 2);
b. Exchange of the phenolate arms’ oxygens for a different heteroatom, e.g., for softer sulfur chalcogenides (Chapter 3);
c. Exchange of the arene anchor for the heteroarene ring 1,3,5-triazine (Chapter 4).
All three strategies involved the synthesis of novel ligands, their coordination to uranium, and the investigation of the complexes’ properties and reactions with a variety of substrates (with a focus on water activation), thus allowing for a comparison of the ligands’ different electronic and steric influences on the reactivity of the uranium center and of the resulting products. Because of the prominent role played by U(IV) and U(V) derivatives in the U(III)-mediated electrocatalytic reduction of water, the syntheses of “benchmark” U(IV) hydroxo (and/or chloride) and U(V) terminal oxo complexes were tackled as well. Structural confirmation of the products was obtained via single-crystal XRD analysis. The isolated compounds were characterized by analytical methods, including NMR, Vis/NIR electronic absorption, IR vibrational, and EPR spectroscopy, as well as elemental analysis and SQUID magnetometry
Engineered injection-extraction systems: sensitivity analysis and implications for groundwater remediation
No full textEffective remediation of contaminated aquifers is often limited by the mixing among reaction partners. An effective method to enhance mixing and reach a faster groundwater remediation is engineered injection extraction (EIE). This technology consists of a sequence of 12-steps of injection and extraction of water through a system of four wells. In this work, we investigate, using Morris sensitivity analysis, which are the most relevant design parameters of EIE among the location of four injection-extraction wells and their pumping rate, in a two-dimensional domain. Moreover, we consider the effect of heterogeneous hydraulic conductivity fields in scenarios of increasing complexity. Our results show that the first steps of the remediation process are the most important ones. The value of the hydraulic conductivity is relevant in case of heterogeneous fields within the treated area, as it generates flow focusing, enhancing plume spreading and mixing. In the case of instantaneous, complete mixing-limited reactions, the sensitivity of the model parameters also depends on the critical mixing ratio, which controls the time needed for the complete degradation of a treatment solution injected in the middle of the EIE system.Open Access funding enabled and organized by Projekt DEAL.Friedrich-Alexander-Universität Erlangen-Nürnberg (1041
A synthetic cohort analysis of postoperative management of primary cardiac angiosarcoma and a case report
No full textBackground Primary cardiac angiosarcoma is a rare and aggressive malignancy originating from the endothelial lining of cardiac blood vessels. The prognosis remains extremely poor. The study was to evaluate postoperative survival in patients with primary cardiac angiosarcoma after treated with adjuvant therapy. Methods A systematic review of PubMed from January 1985 to December 2023 was performed to establish a synthetic cohort of patients undergoing surgery for primary cardiac angiosarcoma. Survival analysis was used to assess the relationship between postoperative adjuvant therapy and prognosis. Univariable and multivariable cox regression analyses were used to identify prognostic factors. We then established and validated a nomogram by receiver operating characteristic (ROC) curves, calibration curves and decision curve analysis (DCA). Moreover, we present a case of 49-year-old patient with primary cardiac angiosarcoma. Results In the synthetic cohort, the patients with postoperative adjuvant therapy reached longer overall survival (OS) and progression-free survival (PFS) than those without postoperative adjuvant therapy (median OS: 14 VS 8 months, HR = 5.62, 95%CI: 1.66-19.08, P<0.001; median PFS: 12 VS 6 months, HR = 2.98, 95%CI: 1.03-8.66, P = 0.007; Log rank test). Radiotherapy (HR = 0.14, 95% CI: 0.04-0.54, P = 0.004) and chemotherapy (HR = 0.03, 95% CI: 0.00-0.27, P = 0.002) were significantly correlated with better OS. DCA and ROC curves confirmed the nomogram can predict postoperative 6-month survival in patients with primary cardiac angiosarcoma. OS was indistinguishable between patients with R0 or R1 resection (10 VS 10 months, HR = 0.99; 95%CI: 0.34-2.86; P = 0.986). However, compared to patients underwent R1 resection, patients undergoing R0 resection have longer but not statistically significant PFS (10 VS 7 months, HR = 2.16; 95%CI: 0.83-5.61; P = 0.090). Conclusion The prognosis of patients with primary cardiac angiosarcoma remains extremely poor, even with surgical resection. Postoperative adjuvant therapy was associated with significantly better survival in a small cohort of patients with primary cardiac angiosarcoma. Further studies are warranted to guide future recommendations. Systematic Review Registration https://www.crd.york.ac.uk/prospero/ , identifier CRD420251139779
Association between the 400-m walk test and sensor-based daily physical activity in frail and sarcopenic older adults
No full textKey summary points Aim Our objective was to examine the association between the time and number of stops during the 400-meter walk test (400MWT) and average daily steps and walking cadence in a German cohort of frail and sarcopenic community-dwelling older adults at first observation (FO) and individual last observation (LO) after at least 11 months. Findings Stops during the 400MWT led to a longer time. Time in the 400MWT was significantly associated with average daily steps and walking cadence at FO and LO (24.5 ± 8.5 months). Stops alone were not associated with average daily steps and walking cadence. Message Gait speed under laboratory conditions can be used to estimate daily physical activity, represented by average daily steps and walking cadence, in frail and sarcopenic older adults.Purpose Physical activity (PA) is recommended for frail and sarcopenic older adults as an essential means of preventing negative health outcomes and decline in functional abilities. Our objective was to examine the association between the time and/or stops during the 400-m walk test (400MWT) and average daily steps and walking cadence in a German cohort of frail and sarcopenic community-dwelling older adults. Methods For this sub-study the German cohort of 104 frail and sarcopenic older adults (i.e., SPRINTT) aged 80.8 ± 5.2 years was divided into participants having made none or one stop or more than one stops, referred to as non-stoppers and multi-stoppers. The characteristics and general health state (physical function, disease state, concerns about falling) at first observation (FO) and individual last observation (LO) after at least 11 months (mean 24.5 ± 8.5 months) were examined. Daily PA represented by average daily steps and walking cadence was assessed over three to seven days at FO and LO using the activPAL3 micro. Time and stops made during the 400MWT and their association with daily PA were investigated using regression with bootstrapping. Results Out of 104 frail and sarcopenic older adults, 84 non-stoppers (female: n = 54; 64.3%) had a median time in 400MWT of 509 s (Inter Quartile Range (IQR) 324–875), a median number of daily steps of 6537 (IQR 1841–19,488) and a median daily walking cadence of 73 steps/minute (IQR 53.3–88.9). 20 multi-stoppers (female: n = 12; 60%) showed a time of 703 s (IQR 479–898), 5642 steps (IQR 2470–11,458) and a cadence of 70.7 steps/minute (IQR 61.4–83.6). Time was significantly associated with average daily steps and walking cadence at both FO and LO, stops alone were not. Conclusion Gait speed under laboratory conditions can be used in clinical settings and research to estimate daily PA, represented by average daily steps and cadence, in frail and sarcopenic older adults.Open Access funding enabled and organized by Projekt DEAL.Friedrich-Alexander-Universität Erlangen-Nürnberg (1041
Patient-specific hiPSC-Podocytes as an in vitro model of genetic FSGS
No full textMutations in podocyte-specific genes are associated with genetic focal segmental glomerulosclerosis (FSGS), yet the potential for targeted treatments remains uncertain. Therefore, patient-specific models are essential for understanding cellular phenotypes, identifying personalized therapies, and avoiding ineffective treatments. This study utilized patient-specific human induced pluripotent stem cell (hiPSC)-Podocytes to investigate cellular phenotypic and functional alterations associated with genetic FSGS in vitro. HiPSC-Podocytes were generated from a patient with a mutation in the inverted formin 2 ( INF2 ) gene, who showed a partial clinical response to steroid treatment. Compared to healthy donor-derived hiPSC-Podocytes, the patient-specific hiPSC-Podocytes exhibited decreased protrusion length, reduced levels of actin-associated markers, and alterations in INF2 protein levels. Additionally, actin filaments were disrupted, characterized by increased actin depolymerization. Next to the actin-modulating agent Bis-T-23, the steroid Solu-Decortin H (SDH) improved the actin cytoskeleton in the patient-specific cells, which aligned with the patient’s partial response to steroids. This underscores the importance of personalized treatment strategies based on specific cellular responses in genetic FSGS.Open Access funding enabled and organized by Projekt DEAL.Universitätsklinikum Erlangen (8546
Investigating firn structure and density in the accumulation area of the Grosser Aletschgletscher using ground-penetrating radar
No full textThe role of firn structure and density in geodetic glacier mass balance estimation has been constrained, with studies in alpine conditions primarily relying on models. Our research focuses on understanding the firn structures, density, and accumulation history in the Grosser Aletschgletscher accumulation area, using field methods mainly involving ground-penetrating radar (GPR) as a geophysical tool and glaciological methods such as snow pits, snow cores, firn cores, and isotope analysis. We characterise the firn structure and determine the spatial firn density–depth profiles by estimating electromagnetic wave velocities using the GPR-based common mid-point (CMP) method. This is done by identifying reflection hyperbolae using semblance analysis of the CMP data set. Three density–depth profiles, up to 37 m depth, were obtained at various locations within the glacier accumulation area. The Ligtenberg (LIG) and Kuipers Munnekee (KM) firn compaction models were selected from the community firn models (CFMs) to evaluate how well the model results matched the observations. These models were predominantly adjusted to fit the estimated 1-D firn density profiles from CMP measurements by optimising model coefficients based on regional Alpine climatic conditions, rather than the conventional method of tuning to the firn core density profiles. Further, a method is introduced to estimate accumulation history by chronologically identifying GPR-derived internal reflection horizons (IRHs) as annual firn layers, by comparing the estimated snow water equivalent (SWE) within each IRH to SWE from long-term point mass balance measurements available at the accumulation area of the glacier. We investigated the spatial distribution of the firn density and the glacier's accumulation history over the past 10–14 years (2010–2023) using a 1.8 km GPR transect, supported by CMP-derived density–depth profiles. Furthermore, our findings emphasise the importance of direct measurements, such as snow cores, firn cores, and isotope samples, in identifying the previous end-of-summer horizon. In this study, we demonstrate the potential of integrating GPR, direct measurements, and firn compaction models to monitor firn structures and density, ultimately enhancing glacier mass balance estimation in future research.Elitenetzwerk Bayer
The Prognostic Role of Para-Aortic Lymph Node Metastasis in Patients with Resected Pancreatic Adenocarcinoma
No full textSimple Summary Pancreatic cancer remains one of the deadliest cancers, and even after surgery, many patients experience early recurrence and poor survival. During surgery for pancreatic cancer, removal and examination of para-aortic lymph nodes can help determine how far the cancer has spread. However, it is unclear whether removing these lymph nodes improves survival. This study examined patients who underwent pancreatic surgery with or without para-aortic lymph node dissection to assess its value. The results showed that removing these lymph nodes does not directly improve survival but helps identify patients whose cancer has spread further and who therefore have a worse prognosis. Recognizing these high-risk patients may help clinicians tailor follow-up treatments, such as intensified chemotherapy, and improve decision-making in patient care in the future.Background: This study aimed to evaluate the prognostic significance of para-aortic lymph node dissection (PALND) during pancreatic head resection and the impact of para-aortic lymph node metastasis (PALN+) on survival outcomes in patients with resected pancreatic ductal adenocarcinoma (PDAC). Methods: A retrospective analysis was conducted on 198 patients who underwent primary pancreatic head resection for PDAC at the University Hospital Erlangen between 2003 and 2022. Patients were stratified based on the presence or absence of PALND and PALN metastases, and their clinicopathological characteristics and survival outcomes were compared. Results: Of the 198 patients, 113 (57%) underwent additional PALND. PALND itself had no significant impact on overall survival (OS) or disease-free survival (DFS) compared to those without PALND. Among patients who underwent PALND, 17 (15%) had PALN metastases (PALN+). PALN+ patients exhibited significantly worse pathological features, including a higher rate of regional lymph node metastases (pN+), lymphovascular invasion (L1) and vascular invasion (V1). Survival analysis showed that PALN+ was associated with significantly poorer OS (8.7 vs. 29.3 months, p < 0.001) and DFS (3.8 vs. 17.0 months, p < 0.001). In multivariate analysis, PALN+ was confirmed as an independent prognostic factor for both OS (HR 1.9 [1.0–3.6], p = 0.035) and DFS (HR 2.2 [1.2–4.0], p = 0.006). Conclusions: While PALND does not impact survival outcomes in PDAC, it plays a crucial role in identifying PALN+ patients, who have significantly worse prognoses. PALN status should be integrated into clinical decision-making, particularly when considering intensified adjuvant therapy.This research received no external funding
Von der optimalen Steuerung zur zufälligen und neuronalen Netzapproximation
No full textThis thesis advances three main research directions—optimal control, random numerical methods, and the expressivity power of narrow, deep neural networks.
Part I establishes exponential turnpike results for two classes of optimal control problems with long time horizons. The turnpike property refers to the phenomenon whereby optimal trajectories remain close, for most of the horizon, to the steady-state solution of the associated static optimal control problem. We first consider a family of linear parabolic equations with coefficients in a suitable class and prove that the turnpike estimate holds uniformly with respect to both time and parameters. The core of the argument is the use of uniform null-controllability estimates that enable the uniform stabilization of the associated Riccati operator. In a second setting, we study finite-dimensional optimal control problems with random coefficients, where the state is observed in expectation. Under suitable average-stabilizability and detectability conditions, we prove that the expected optimal trajectory satisfies the exponential turnpike property with respect to its stationary state, while the variance decays at the same rate—thus extending the turnpike principle to ensemble optimal control under uncertainty.
Part II presents a systematic analysis of the Random Batch Method (RBM), an algorithm originally developed to reduce the computational cost of interacting particle systems by approximating full interactions through random subsampling. The first setting introduces a discretize-then-randomize approach for approximating partial differential equations (PDEs), with the observation that, in the case of PDEs on graphs, this corresponds to a random domain decomposition in which the equation is solved alternately on randomly selected subgraphs. This idea is then extended to a continuous RBM framework on graphs, formulated directly at the level of the PDE, where convergence is established independently of any underlying discretization. The method is further generalized to gradient flows in both finite- and infinite-dimensional Hilbert spaces, with convergence of the batch-based dynamics to the exact flow shown under suitable assumptions. We then analyze the application of RBM to nonlinear ordinary differential equations, where the vector field is expressed as a sum of nonlinear components. This yields a convergence result for the RBM approximation of nonlinear dynamics, which is subsequently used to derive a random batch scheme for the associated continuity equation and for optimal control problems governed by nonlinear ODEs. This setting is particularly relevant for continuous-time models in machine learning, such as Neural ODEs, which arise as infinite-depth limits of residual neural networks (ResNets); in this context, RBM provides a rigorous interpretation of continuous-time dropout, a technique that randomly deactivates neurons during training. Throughout, numerical experiments are provided to illustrate the convergence behavior of the proposed methods and to demonstrate the computational savings achieved through random batching.
Part III focuses on the expressivity of neural networks, addressing their capacity to approximate functions or interpolate arbitrary datasets. We consider multilayer perceptrons (MLPs) with ReLU activations, which—unlike ResNets—cannot be interpreted as discretizations of continuous dynamical systems. The first set of results concerns finite-sample memorization, also known as universal interpolation: given any finite dataset, MLPs are capable of mapping inputs exactly to their prescribed labels. From a control-theoretic perspective, these results can be viewed as establishing the simultaneous, or ensemble, controllability of the corresponding discrete-time dynamical system. We prove that MLPs of width two satisfy the universal interpolation property and provide a constructive proof that reveals the internal mechanism by which successive layers build up the interpolant. This constructive approach further yields an explicit estimate on the number of layers required to achieve exact interpolation, and estimation of the norm of the constructed parameters. We then explore the implications of our constructive interpolation scheme for supervised learning with regularization. Although our networks are derived independently of any optimization procedure, we leverage the explicit construction to derive upper bounds on the regularized empirical loss. In particular, we show that the norm of the minimizer of the regularized loss is uniformly controlled by the norm of our interpolating solution. As the regularization parameter vanishes, the optimizer converges to a minimal-norm interpolant. Our last contribution addresses the classical universal approximation theorem—a density result in function spaces—within families of neural network architectures. We show that ReLU MLPs of fixed width, equal to the input dimension plus one, can approximate any integrable function on a bounded domain. This result is again established constructively, providing insight into the underlying approximation strategy. In addition, we derive quantitative estimates on the required depth in terms of the Sobolev regularity of the target function, yielding explicit bounds that highlight the expressive power of narrow but sufficiently deep networks.Diese Dissertation behandelt drei zentrale Forschungsfragen: die Optimalsteuerung, zufällige numerische Verfahren und das Darstellungsvermögen schmaler, tiefer neuronaler Netze.
Teil I etabliert exponentielle Turnpike-Resultate für zwei Klassen optimaler Steuerungsprobleme mit langen Zeithorizonten. Die Turnpike-Eigenschaft bezeichnet das Phänomen, dass optimale Trajektorien über weite Teile des Horizonts nahe an der stationären Lösung des zugehörigen statischen Optimalsteuerungsproblems verbleiben. Hierzu betrachten wir zunächst eine Familie linearer parabolischer Gleichungen mit Koeffizienten aus einer geeigneten Klasse und zeigen, dass die Turnpike-Abschätzung sowohl zeit- als auch parameterunabhängig gilt. Kern des Arguments ist der Einsatz uniformer Null-Kontrollierbarkeitsabschätzungen, die eine gleichmäßige Stabilisierung des zugehörigen Riccati-Operators erlauben. Als zweiten Beitrag untersuchen wir endlich-dimensionale Optimalsteuerungsprobleme mit
zu-fälligen Koeffizienten, bei denen der Systemzustand im Erwartungswert beobachtet wird. Unter geeigneten Bedingungen zur mittleren Stabilisierbarkeit und Detektierbarkeit beweisen wir, dass die erwartete optimale-Trajektorie die exponentielle Turnpike-Eigenschaft bezüglich ihres stationären Zustands erfüllt, während die Varianz mit derselben Rate abnimmt – womit das Turnpike-Prinzip auf die Ensemble-Optimalsteuerung unter Unsicherheit verallgemeinert wird.
Teil II präsentiert eine systematische Analyse der Random Batch Method (RBM), eines Verfahrens zur Reduktion des Rechenaufwands in wechselwirkenden Teilchensystemen durch zufälliges Subsampling der Wechselwirkungen. Im ersten Rahmen führen wir einen ``discretize-then-randomize''-Ansatz zur Approximation partieller Differentialgleichungen (PDEs) ein und zeigen, dass dies im Fall von PDEs auf Graphen einer zufälligen Gebietszerlegung entspricht, bei der die Gleichung abwechselnd auf zufällig ausgewählten Subgraphen gelöst wird. Diese Idee wird auf einen kontinuierlichen RBM-Ansatz auf Graphen erweitert, der direkt auf PDE-Ebene formuliert ist und dessen Konvergenz unabhängig von einer zugrundeliegenden Diskretisierung nachgewiesen wird. Das Verfahren wird weiter auf Gradientenflüsse in endlich- und unendlich-dimensionalen Hilberträumen verallgemeinert, wobei unter geeigneten Voraussetzungen die Konvergenz der batchbasierten Dynamik zum exakten Fluss gezeigt wird. Anschließend analysieren wir RBM für nichtlineare gewöhnliche Differentialgleichungen (ODE), deren Vektorfeld als Summe nichtlinearer Komponenten dargestellt wird. Dies führt zu einem Konvergenzresultat für die RBM-Approximation der nichtlinearen Dynamik, das wiederum zur Herleitung eines Random-Batch-Schemas für die zugehörige Kontinuitätsgleichung und für Optimalsteuerungsprobleme unter nichtlinearen ODEs genutzt wird. Dieses Setting ist besonders relevant für kontinuierliche Modelle im maschinellen Lernen wie Neural-ODEs, die als unendlich-tiefe Grenzfälle von Residual-Netzen (ResNets) entstehen; hier bietet RBM eine rigorose Interpretation des kontinuierlichen Dropouts, bei dem Neuronen zufällig deaktiviert werden. Numerische Experimente illustrieren durchgängig das Konvergenzverhalten der vorgeschlagenen Methoden und demonstrieren die erzielten Rechenzeiteinsparungen.
Teil III widmet sich dem Darstellungsvermögen neuronaler Netze, insbesondere ihrer Fähig-keit, Funktionen zu approximieren oder beliebige Datensätze zu interpolieren. Wir betrachten mehrschichtige Perzeptronen (MLPs) mit ReLU-Aktivierungen, die – im Gegensatz zu ResNets – nicht als Diskretisierungen kontinuierlicher dynamischer Systeme interpretiert werden können. Zunächst behandeln wir das Erinnerungsvermögen endlicher Datenmengen, auch universelle Interpolation genannt: Gegeben eine endliche Stichprobe, ermöglichen MLPs eine exakte Zuordnung der Eingaben zu vorgegebenen Labels. Aus kontrolltheoretischer Sicht entsprechen diese Ergebnisse der gleichzeitigen bzw. Ensemble-Steuerbarkeit des zugrunde liegenden diskreten dynamischen Systems. Wir zeigen, dass MLPs der Breite zwei die universelle Interpolationseigenschaft erfüllen, und liefern einen konstruktiven Beweis, der den internen Mechanismus offenlegt, mit dem aufeinanderfolgende Schichten den Interpolanten aufbauen. Dieser konstruktive Zugang ermöglicht zudem eine explizite Abschätzung der benötigten Schichtzahl für die exakte Interpolation. Wir untersuchen anschließend die Auswirkungen unseres konstruktiven Interpolationsverfahrens im Kontext des überwachten Lernens mit Regularisierung. Obwohl unsere Netze unabhängig von einem Optimierungsverfahren konstruiert wurden, nutzen wir die explizite Konstruktion, um obere Schranken für den regularisierten empirischen Fehler abzuleiten. Insbesondere zeigen wir, dass die Norm des Minimierers der regularisierten Zielfunktion gleichmäßig durch die Norm unserer Interpolationslösung kontrolliert wird. Im Grenzfall verschwindender Regularisierung konvergiert der Optimierer gegen ein Interpolationsnetzwerk mit minimaler Norm. Unser zweiter Beitrag befasst sich mit dem klassischen universellen Approximationssatz – einer Dichtheitsaussage in Funktionräumen – für Familien von neuronalen Netzen. Wir beweisen, dass ReLU-MLPs fester Breite, gleich der Eingabedimension plus eins, jede integrierbare Funktion auf einem beschränkten Gebiet approximieren können. Auch dieser Satz wird konstruktiv hergeleitet und liefert Einblicke in die zugrunde liegende Approximationsstrategie. Überdies leiten wir quantitative Abschätzungen zur erforderlichen Netztiefe in Abhängigkeit von der Sobolev-Regularität der Zielfunktion ab, wodurch enge, aber hinreichend tiefe Architekturen hinsichtlich ihres Darstellungsvermögens konkret bewertet werden können
Ultrathin [C1C1Im][Tf2N] Layers on Supported Mn3O4(001) Films
No full textIn the Solid Catalyst with Ionic Liquid Layer concept, activity and selectivity of a solid catalyst such as metal particles supported on porous oxides can be tuned and optimized by ultrathin ionic liquid (IL) coatings. In this context of ILs interacting with oxide surfaces, we report on the adsorption behavior and thermal stability of the IL 1,3-dimethylimidazolium bis(trifluoromethylsulfonyl)imide ([C1C1Im][Tf2N]) on well-ordered Mn3O4(001) thin films grown on Au(111) using angle-resolved X-ray photoelectron spectroscopy. We observe the formation of a two dimensional IL wetting layer for IL coverages ≤ 0.5 ML, while multilayers on top of the closed wetting layer (≥ 0.5 ML) grow in 3D islands. For both coverage regimes, the IL anion is preferentially oriented in a cis conformation with the CF3 groups preferentially pointing to the vacuum and SO2 groups towards the oxide surface. Temperature-programmed XPS shows that the multilayer starts to desorb at around 300 K as neutral ion pairs. Above 400 K, the IL wetting layer decomposes, which is characterized by the disappearance of cation-related XPS signals until ~ 450 K, while anion signals remain until 500 K. In comparison to [C1C1Im][Tf2N] films on reactive metal surfaces such as Cu(111) and Pt(111), the IL exhibits a higher thermal stability on Mn3O4(001).Graphical abstractOpen Access funding enabled and organized by Projekt DEAL.Friedrich-Alexander-Universität Erlangen-Nürnberg (1041
Molecular Characterization of the GALC Mutation Thr112Ala Causing Krabbe Disease
No full textKrabbe disease is a rare and severe lysosomal disorder affecting the white matter of the central and peripheral nervous system. It is characterized by neurodegeneration, with the most common form being infantile Krabbe disease, typically diagnosed within the first year of life. This autosomal-recessive disease is caused by mutations in the GALC gene, which encodes the lysosomal enzyme β-galactocerebrosidase. This study focuses on a β-galactocerebrosidase variant, with Thr112Ala identified as a homozygous mutation in a patient with infantile Krabbe disease. To understand the structural effects of this mutation, we conducted all-atom molecular dynamics simulations of both the mutant and wild-type (wt) enzymes at cytosolic (pH 7.0) and lysosomal pH (pH 4.5), as β-galactocerebrosidase is localized in the lysosome. The results showed differences in protein flexibility, the hydrogen bond network, and the stability of secondary structure elements between the wild type and mutant enzymes. Additionally, the mutation affected the size of the substrate-binding pocket at lysosomal pH, even though the mutation site is not part of the active/binding site of the enzyme. These findings provide valuable insights into how the mutation impacts the structure of β-galactocerebrosidase in the lysosomal environment, contributing to the understanding of Krabbe disease’s molecular mechanisms.E.S. received funding from Universitätsbund Erlangen-Nürnberg e.V.Universitätsbund Erlangen-Nürnberg e.V